For business scenarios such as content classification, customer service Q&A, information extraction, and code completion, open-source models can already deliver sufficiently good results at a fraction of the cost of closed-source solutions.